Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -70,6 +70,9 @@ Evaluation results:
|
|
70 |
| ----- | ----- |
|
71 |
| openai/whisper-large-v3-turbo | 14.14% |
|
72 |
| openai/whisper-large-v3 | 13.79% |
|
|
|
|
|
|
|
73 |
|
74 |
openai/whisper-large-v3-turbo
|
75 |
|
@@ -174,8 +177,7 @@ for i in tqdm(range(0, len(test_data), batch_size)):
|
|
174 |
transcriptions = processor.batch_decode(generated_ids, skip_special_tokens=True)
|
175 |
references.extend(batch["transkription"])
|
176 |
predictions.extend(transcriptions)
|
177 |
-
|
178 |
-
print("WER: ", wer(references, predictions, reference_transform=wer_standardize_contiguous, hypothesis_transform=wer_standardize_contiguous)*100)
|
179 |
print("*"*20)
|
180 |
print("WER: ", wer(references, predictions, reference_transform=wer_standardize_contiguous, hypothesis_transform=wer_standardize_contiguous)*100)
|
181 |
print("*"*20)
|
|
|
70 |
| ----- | ----- |
|
71 |
| openai/whisper-large-v3-turbo | 14.14% |
|
72 |
| openai/whisper-large-v3 | 13.79% |
|
73 |
+
| primeline/whisper-large-v3-german | 10.54% |
|
74 |
+
| nyrahealth/CrisperWhisper | 8.519036731165182 |
|
75 |
+
| primeline/whisper-large-v3-turbo-german | 4.77% |
|
76 |
|
77 |
openai/whisper-large-v3-turbo
|
78 |
|
|
|
177 |
transcriptions = processor.batch_decode(generated_ids, skip_special_tokens=True)
|
178 |
references.extend(batch["transkription"])
|
179 |
predictions.extend(transcriptions)
|
180 |
+
|
|
|
181 |
print("*"*20)
|
182 |
print("WER: ", wer(references, predictions, reference_transform=wer_standardize_contiguous, hypothesis_transform=wer_standardize_contiguous)*100)
|
183 |
print("*"*20)
|